Health Value Pipelines are threads of value woven into the healthcare ecosystem by way of data processing and information creation, both by humans and intelligent automations, and by collaborations between the two.
We work with companies that are improving patient care or reducing societal healthcare costs, to upgrade their health value pipelines by building and commoditizing software components that allow humans and computers to more efficiently and empathetically provide healthcare services.
Health Pipelines are collaborative human-expert/computer-automation processing pipelines at multiple levels of abstraction for health care data and services.
In modern large-scale software development there is significant adoption of data processing pipelines. Deriving from earlier scaling methodologies like MapReduce these are queue and event based architectures that handle huge flows of information in realtime - think Facebook servers processing an incoming click stream through layers of storage, aggregation, processing and data mining.
Health data pipelines handle streams of events like patient admit, discharge or transfer information; order queues and results feeds; referrals and appointments; and many other data flows. Healthcare systems are very heterogeneous at all levels of the communications stack, so basic connectivity, ingestion and data normalization are challenging if starting from scratch.
Building on Data Pipelines, Health Information Pipelines process the raw data to derive healthcare information. These are normalized, time sequenced, de-duplicated items of interest in the delivery of healthcare.
Building on Information Pipelines, Health Value Pipelines provide a full end to end thread of healthcare value-add. Threads might include
- Tracking a prescription order, reminding the patient if it's not picked up, and notifying care team members of progress.
- Getting patient approval, uploading a chart, evaluating the patient against applicable on-going trials, summarizing chart contents and trial options, and passing that information to a review panel.
Health care is an intrinsically human to human activity. Whats more human than caring for others?
Healthcare is a data-rich environment. New ML tools change the equation on what only-humans can do.
It might not seem like it but we are approaching a golden age of Healthcare IT, at least in the US.
- At the macro level no matter what happens, business models and value propositions are changing and will continue to change. Managed care, increased specialization and cost pressures are allowing smaller companies to enter the market in targeted niches and then seek to disrupt.
- 10+ years after the HITECH Act healthcare data is finally becoming readily available. There are government mandates for access by patients and for exchange among care providers. Its true that no two EHR's or healthsystems generate compatible datasets but the data is starting to flow nonetheless.
- Healthcare data is HUGE and only getter bigger with the incorporation of DNA, personal tracking data and social determinants of health.
- Over the last few years advances in machine learning have made significant new tools available for natural language processing, pattern recognition and computer aided decision making. Applying these tools to large healthcare data sets has the potential to significantly improve care outcomes while simultaneously automating away work and reducing costs.
- The FHIR standard shows great potential to unify the world of the internet and mobile software with healthcare; and its success and adoption are encouraging further standardization efforts (eg mCODE ).
Based on these dynamics, more and more mission-driven healthcare companies are forming to solve targeted problems by building and delivering new healthcare value pipelines. In doing so they are often tackling common problems to build out needed pipeline components. Insofar as these components are not core to a companies intrinsic value-add it makes sense to outsource and/or re-use pre-existing building blocks.